Interesting point.

However I think that good developers who are on the lookout for it can flag constructs which they understand but they think others might not. Furthermore good developers are more apt to notice the maintainance implications of seemingly innocuous constructs. Therefore if you have good developers who are concious of these points, and who have exposure to what others know (eg through assisting in training, answering questions, etc), I think they can take into account the bottom half in considering maintainability for code reviews.

But still I have to wonder at the value of doing a modified usability study to find maintainability implications. Take code from programmer A, hand it to programmer B, and have experienced observer C take notes as B talks through trying to figure out A's code...

UPDATE
Inserted an important "not" that Corion caught.